Moment-based estimation of switching activity for correlated distributions

Energy estimation at high-levels of abstraction is getting a mandatory step in current design flows. With that goal, a moment-based stochastic technique has been recently introduced [5,4]. It provides fast power estimations using only word-level parameters; however, in its current formulation it is limited to un-correlated scenarios. Therefore, it cannot be broadly applied. In this paper an extension is proposed to suppress the aforementioned limitation. The technique employs the characterization of the probability density function (PDF) with a projection in a two-dimensional base constructed upon a novel set of orthogonal polynomials. The approach has been validated with practical experiments. Comparisons with reference bit level simulations and previous works are reported to assess the accuracy of the technique.